1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227
|
// This is a mildly modified version of the code in SciPy's
// scipy.optimize.linear_sum_assignment, extracted from
// rectangular_lsap.cpp.
// https://github.com/scipy/scipy/blob/main/scipy/optimize/rectangular_lsap/rectangular_lsap.cpp
// As such it is subject to the following notice:
/*
Copyright (c) 2001-2002 Enthought, Inc. 2003-2023, SciPy Developers.
All rights reserved.
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions
are met:
1. Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
2. Redistributions in binary form must reproduce the above
copyright notice, this list of conditions and the following
disclaimer in the documentation and/or other materials provided
with the distribution.
3. Neither the name of the copyright holder nor the names of its
contributors may be used to endorse or promote products derived
from this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
"AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
This code implements the shortest augmenting path algorithm for the
rectangular assignment problem. This implementation is based on the
pseudocode described in pages 1685-1686 of:
DF Crouse. On implementing 2D rectangular assignment algorithms.
IEEE Transactions on Aerospace and Electronic Systems
52(4):1679-1696, August 2016
doi: 10.1109/TAES.2016.140952
Author: PM Larsen
*/
#include <algorithm>
#include <iostream>
#include <limits>
#include <numeric>
#include <vector>
namespace RDKit {
namespace RascalMCES {
template <typename T>
std::vector<size_t> argsortIter(const std::vector<T> &v) {
std::vector<size_t> index(v.size());
std::iota(index.begin(), index.end(), 0);
std::sort(index.begin(), index.end(),
[&v](size_t i, size_t j) { return v[i] < v[j]; });
return index;
}
static int augmentingPath(size_t nc, std::vector<int> &cost,
std::vector<double> &u, std::vector<double> &v,
std::vector<size_t> &path,
std::vector<size_t> &row4col,
std::vector<double> &shortestPathCosts, size_t i,
std::vector<bool> &SR, std::vector<bool> &SC,
std::vector<size_t> &remaining, double *p_minVal) {
double minVal = 0;
// Crouse's pseudocode uses set complements to keep track of remaining
// nodes. Here we use a vector, as it is more efficient in C++.
size_t numRemaining = nc;
for (size_t it = 0; it < nc; it++) {
// Filling this up in reverse order ensures that the solution of a
// constant cost matrix is the identity matrix (c.f. #11602).
remaining[it] = nc - it - 1;
}
std::fill(SR.begin(), SR.end(), false);
std::fill(SC.begin(), SC.end(), false);
std::fill(shortestPathCosts.begin(), shortestPathCosts.end(),
std::numeric_limits<double>::max());
// find shortest augmenting path
int sink = -1;
while (sink == -1) {
// Clearly this will produce an overflow and set index to a large integer.
// It is how the original code did it, and I assume whoever wrote it knew
// what they were doing.
size_t index = -1;
double lowest = std::numeric_limits<double>::max();
SR[i] = true;
for (size_t it = 0; it < numRemaining; it++) {
size_t j = remaining[it];
double r = minVal + cost[i * nc + j] - u[i] - v[j];
if (r < shortestPathCosts[j]) {
path[j] = i;
shortestPathCosts[j] = r;
}
// When multiple nodes have the minimum cost, we select one which
// gives us a new sink node. This is particularly important for
// integer cost matrices with small co-efficients.
if (shortestPathCosts[j] < lowest ||
(shortestPathCosts[j] == lowest &&
row4col[j] == static_cast<size_t>(-1))) {
lowest = shortestPathCosts[j];
index = it;
}
}
minVal = lowest;
if (minVal ==
std::numeric_limits<double>::max()) { // infeasible cost matrix
return -1;
}
size_t j = remaining[index];
if (row4col[j] == static_cast<size_t>(-1)) {
sink = j;
} else {
i = row4col[j];
}
SC[j] = true;
remaining[index] = remaining[--numRemaining];
}
*p_minVal = minVal;
return sink;
}
int lapMaximize(const std::vector<std::vector<int>> &costsMat,
std::vector<size_t> &a, std::vector<size_t> &b) {
if (costsMat.empty() || costsMat.front().empty()) {
return 0;
}
size_t nr = costsMat.size();
size_t nc = costsMat.front().size();
bool transpose = nc < nr;
std::vector<int> cost(nc * nr);
// for maximization, take -ve of costs.
for (size_t i = 0; i < nr; ++i) {
for (size_t j = 0; j < nc; ++j) {
if (transpose) {
cost[j * nr + i] = -costsMat[i][j];
} else {
cost[i * nc + j] = -costsMat[i][j];
}
}
}
if (transpose) {
std::swap(nc, nr);
}
// initialize variables
std::vector<double> u(nr, 0);
std::vector<double> v(nc, 0);
std::vector<double> shortestPathCosts(nc);
std::vector<size_t> path(nc, -1);
std::vector<size_t> col4row(nr, -1);
std::vector<size_t> row4col(nc, -1);
std::vector<bool> SR(nr);
std::vector<bool> SC(nc);
std::vector<size_t> remaining(nc);
// iteratively build the solution
for (size_t curRow = 0; curRow < nr; curRow++) {
double minVal;
int sink = augmentingPath(nc, cost, u, v, path, row4col, shortestPathCosts,
curRow, SR, SC, remaining, &minVal);
if (sink < 0) {
return -1;
}
// update dual variables
u[curRow] += minVal;
for (size_t i = 0; i < nr; i++) {
if (SR[i] && i != curRow) {
u[i] += minVal - shortestPathCosts[col4row[i]];
}
}
for (size_t j = 0; j < nc; j++) {
if (SC[j]) {
v[j] -= minVal - shortestPathCosts[j];
}
}
// augment previous solution
size_t j = sink;
while (1) {
size_t i = path[j];
row4col[j] = i;
std::swap(col4row[i], j);
if (i == curRow) {
break;
}
}
}
if (transpose) {
size_t i = 0;
for (auto v : argsortIter(col4row)) {
a[i] = col4row[v];
b[i] = v;
i++;
}
} else {
for (size_t i = 0; i < nr; i++) {
a[i] = i;
b[i] = col4row[i];
}
}
return 0;
}
} // namespace RascalMCES
} // namespace RDKit
|